David, I'll post step by step explanation for sundoku and extraregions, which were not so difficult, but they are some steps that are not so easy to find, especially under time pressure.




It's nice to see that there is no unanimity about difficulty of some grids . When I tested Double sum, I must say I found it very hard, and I could not find a strategy to solve it in a good time. I think all testers found it hard (best of us solved it in 17'), this is why it had so many points.

Fred

To continue my mini discussion with Bastien and Fred - another thing to take into account is that each puzzle author inevitably has his/her own style. I imagine that as a group you are well accustomed to each other's puzzles. Of course you also have the extra factors of time-pressure, printing time, answer-entry time and all the rest as you rightly say.

Very strange re the double sum. During the test itself I must have spent that last ~45 minutes on the 0-8 killer. I got about a third to a half way in, thinking to myself this will now fall very quickly, like most killers do when you get a few digits placed. Except I didn't. In comparison, solving afterwards, the double sum was quite gentle. It definitely didn't take longer than 10 minutes and was probably closer to 5. My explanation is that the 0-8 killer didn't feel very constrained at all (I'm probably missing something), and the double sum most certainly did - to the point where I thought I'd gone wrong a couple of times only to spot that actually I had made the sum elsewhere in the row/column.

I'm glad I'm in good company with the sudokurve. I thought I had it solved, but as I was entering I saw the mistake. I then decided to put it aside, and thanks to the 0-8 killer never managed to get back round to it!

This problem is fixed now.





Attachments
----------------
text3428.png (6KB - 4 downloads)

I enjoyed this test. Best sudoku finish so far for me. But that is probably partly because it was only a small set of puzzles. I generally fair better with a smaller number of harder sudoku puzzles than a lot of easier sudoku puzzles.

My favourites during competition were the extra regions and 0-8 killer sudoku. I didn't need to guess for the extra regions, found the logic path quite nicely.
I spent a bit too much time on the Jigsaw Wordoku as I had at first missed the "repeated letters can't be adjacent" rule. I got pretty far with it till I realised that one.
The first time I did the Sudokurve I messed it up too, till I noticed 2 6's in a long bent row. I found the outside lines a bit too hard to follow.
The Sundoku was tricky and I messed it up twice in competition. I thought the opening was really nice, just too bad I kept breaking it.
The mixed was also very nice, although I missed the opening during competition. The inequalities did feel a little as an afterthought, as in we set up this really nice thermometer/arrows path except it's not unique yet. I thought the double sum was very hard and when solving it afterwards it didn't really feel like a solid logic path, always felt like I was doing it a bit intuively, constantly finding things that work and assuming i'm still on the right path, not being 100% sure my deductions were right.

All in all, thanks for a fun test. Now if only I can not make an entry error next time. Dumb 4.

About extraregions:



We can easily place every 9 (region 2 & 9, then 5, 4 and finally 6):



Then, we see pairs 38 and 27 in region 9:



That leads to other easy placement:



We now can view a pair 35 in the extra yellow region:



What about the cell R9C6? We can have 1 or 2... but the 2 doesn't fit, otherwise we couldn't place a 2 in blue extra region. So it's a 1 !
And the 35 pair implies a pair 34 in region 1 (by the placement of the 3 in column 3), and of course then a 16 pair in region 1 and another pair 34 in row 3. Then we can place the 5 in region 3. Here we are:



There is only one possible placement for the 1 in blue extraregion, that leads to:






What about the 7 in R6C3? impossible, otherwise there will be two 7's in the red extra region !

So it's a 2 !

Further, looking at placement of the 1 in red region leads to place 1 and 4 in region 6 !

Another extraregion analysis: 5 in Row 7 is in the green region, so there is only one possibility for the 5 in column 4: R2C4. Which leads to:



On R7C1, there is only one possibilty: the 1. Then only one possibility for the 1 in green region: R6C4. Then a few easy placements:



3 is in red region in region 1, in blue region in region2, so the only placement in region 5 is R6C5, which leads to:



We can then place the 5 in region 5, then 4 in region 5 and some easy placements:



Place the 3 in green region, then all the rest are easy placements

Solution:



Fred

I think I could have solved all of the puzzles logically, had I printed them all out and paper solved them as it appears that many people did this. I do not like contests where there is a puzzle which is made so much simpler if you can note take, and becomes a memory puzzle if you can not or an exercise in locating each key on your keyboard when you solve it online. I think the puzzle testers should have tested these puzzles using the logic masters interface, and I'm sure you guys would have realized the competitive advantage that all of the paper solvers had over anyone who did all of the puzzles online.

In the future, if contests are going to contain puzzles with built in advantages to solving it with one means over another, then the scores should be separate and based on how people solve it, or the scores should be adjusted with a multiplier for solving it one way over another. If it's not done this way, I see no reason to compete in any contests hosted on this site because it is not worth the aggravation of having to jump through hoops which many other competitors don't have to just because they printed out the puzzle.

I think LMI has done an excellent job of accommodating solvers throughout the many tests they have run, and have done their best with the sudoku tests to allow for both online and paper solving. Everyone has the option to enter both ways (or to even solve some online and some on paper if that is optimal).

Each method has its pros and cons. Solving online is often much quicker than writing for easy puzzles, avoids the loss of printing time as well as answer entry time and is also free from the transcription errors that come from typing in a solution from paper (which happen at a not insignificant rate - my US teammate MellowMelon seems to have lost two puzzles to simple typos this go around and I've certainly lost credit in the past for mistakes that were not puzzle-solving mistakes). Paper offers the flexibility of any style of notes the solver wants, and is perhaps best suited for very hard puzzles where more notes or possibly guessing become required. I don't see a fair way to weight the two approaches since it is impossible to compare these times rigorously, particularly on novel variations of sudoku.

Reading through the various comments, and testing the interface myself, I can see that online solving was not ideal for three puzzles this time around, once due to the interface and twice due to the puzzle presentation. The 0-8 killer has an issue of the first candidate (0 this time) colliding with cage values. LMI seems to realize this problem and is experimenting with solutions, so future tests should be better in this regard. The Alpha Frame and Double Sum both did something I dislike in sudoku contests (as I commented before in the Numerologidoku on the January test) of requiring extra conversion steps to get to the real data. You are absolutely right that not being able to replace the letters with fixed numbers (as paper solvers could do) is a disadvantage, so in my opinion the authors should have kept the PRINTEMPS theme but simply also printed P=16, R=18, I=9, ... along the rows and columns to save everyone from doing this extra(neous) step.

Where I don't agree with you is that the Jigsaw Wordoku was unfair for online solvers. If your complaint is "I don't know where the letters are on my keyboard", then let's admit that that is not a fault of the interface and more a variable from your specific skill at typing. Yes the numpad, and remembering 1-9 complete sets, is helpful for speedy sudoku solving, but I see no fault in going to letters either from the authors or the site administrators. Letters trips up paper solvers too as we can't fall back on 1-9 sets either.

Hearing your perspective on online solving, let me share mine from the paper side. As I and others have commented, paper had its own disadvantages this time around. The color printing seemed to take a fair bit longer than usual for these tests. For some, the PDF literally took forever to print. So instantly disqualifying yourself, or at least losing 2-3 minutes, was a likely consequence. I was 20 seconds from finishing the test and lost 3 minutes to printing. Oh well, goes with the territory. The dark green color on my printout was very difficult to write on and then read from. On both the Extra Region and the Mixed Sudoku, this caused difficulties in solving that were not necessary and also weren't encountered online. Am I threatening to quit because they used dark green and made a poor PDF file? No, but I'll point out the shortcomings so that future tests get better for paper solvers. The new entry system was a huge step forward in fairness, and everything I've ever seen has suggested the LMI organizers bend over backwards to address concerns and improve their tests. As a whole, this test did feel easier on paper, but this has not always been the case and that's just part of the competition.

So I'll agree with you in part but not in whole that there are specific disadvantages to how one goes about solving these contests. But there is no real reason or ability to try to separate online and paper solvers given the goals here of producing world championship level tests. The simplest and safest way to always accommodate online solvers is to simply do classic sudoku of easy to moderate difficulty where the applet will always well match the puzzle. There are hundreds of sites that do that online already and the competition is pretty good. If this is what interests you, please compete at those places. What LMI offers is both challenging tests, new variations, beautiful hand-crafted puzzles, and the challenge of facing the world's best competitors. I'm glad for the rich variety of puzzles we've seen here and, just like all other solvers, I make a choice to solve the test one way or the other each time it happens, and post a score under those same rules.

Edited by motris 2011-03-29 2:43 AM

David,
I definitely understand your frustration for not being able to solve the last 2 Sudokus with 35+ minutes in hand. As one of the LMI organisers (and as someone who has mostly designed the flash interface for Sudoku solving), I'll be the first person to admit that LMI Sudoku interface is definitely not the best around the internet. I'm certain that some (or many) other sites provide much more user-friendly (and powerful) interfaces. But LMI contests differ from many others with the kind of Sudokus we provide. That has been our focus. We have spent more time on making sure that quality of Sudokus is one of the best, rather than fiddling with the flash interface.

At the same time, we invite both kind of players (paper / online) to compete, and many times the pros and cons for both modes balance out. We also have seen players using both the modes in the same test (i.e. solving 6X6 grids or easy 9X9 grids online, while solving other grids on paper).

As many players pointed out, 0-8 had a problem solving online. We probably should have made a note in the forum before the contest. In other Sudokus you've mentioned, paper solves might have certain advantage. But if we decide to only have Sudokus which are fair to paper and online solvers, it will be a huge constraint for authors. Many authors would agree that it limits their creativity, and we probably shouldn't do that.

Please be assured that we take each and every input that we receive and try to work for the best solution. And we'll definitely advise our future authors to keep in mind that a certain percentage of our solvers are online solvers.

Hope to see you participate in future and make LMI tests more competitive.

Deb

Regarding pdf printing problem this time, before the announcement itself, all of us (Fred, Bastien and me) were concerned more about how the pdf will look in a B&W printer.
I printed the pdf (in a high-end HP LaserJet B&W printer) and it came out without any issues. I didn't think much about the color printing.

Frankly, I'm against using colors in puzzle grids at LMI competitions, but we made an exception in this case because of the associated theme. This is the first time we had colors in grids (even MellowMelon had to change his puzzle presentation to accommodate B&W printing).
We are unlikely to have colors in puzzles frequently, but whenever we have we'll ensure that it prints equally well in B&W and color printers. [ Oh, well, at some point in time we discussed to have 2 booklets, one in B&W and one in color ]

Regarding pdf not printing at all in some printers, I've no idea what could be the problem. But Tom C's explanation on UKPA forum seems most reasonable.

A big thank you to Fred for posting a solution path for the Extra Regions Sudoku. Beautifully done and I must apologise for my earlier comments. I have also found a simple logical solution to Sundoku and a logical (but very difficult!) solution to the 0-8 Killer. As predicted, once the time pressure was off, my brain started to work a little better.

This is the second time I partecipate to a contest.
I prefere to solve online and, as requested, i propose some suggestions.
The candidate are very small and removing them with the mouse (ctrl+click) is not so esay.
Is it possible to make the grid resizable with ctrl+mouse wheel?
I discovered today trying to solve the grids I could'nt solve in 2 hours that you can undo all the inputs.
this could encourage the tentatives,
In my opinion it would be better to limit the undos to just few steps.
Great test, anyway.

I agree that the candidates are small. And using ctrl+click is not the best way to remove a candidate...definitely not the quickest and is click-error-prone, i.e if you release the ctrl early you have to re-enter the candidates. I have not used either Undo or Digit colour feature so far. Now, that I know about it, it seems to be a really good feature and can speed up online solving a bit. I don't think there is a need to limit the number of undo steps. Can it be used for trial and error (T&E)? - Yes. But, probably, that is not a big enough reason for restricting the usage. All said, it is a good feature and can only improve the online solving experience, if used well.

Hi David. I am glad you could finally solve the killer logically. The puzzle wasn't that hard if you managed to follow the intended path, but of course this path wasn't particularly easy to spot (well, the puzzle was worth 140pts...).

Time for a global look on the test.

Generally, puzzles were widely appreciated and this is always a major satisfaction, being the main goal of each and every author. More of that, some of them got some emphatic comments, in particular Mixed Sudoku from Laurent Pierre (the man who depicts himself as a "beginner author") ; we can only be satisfied of that and it fully rewards the work which has been done.

My own 0-8 Killer also got some very nice comments and this is a great reward. As I said to Serkan on saturday, this puzzle caused me some torment. My original idea was to simply create a basic killer with 20/3 areas but after several tries, I couldn't get a satisfying puzzle in terms of difficulty. Then came the idea of using a different set of digits, and the "0-8" set appeared to offer lots more possibilities to play with - in particular, the sum "3" now admitted 3 different combinations, including a 3-cells one.
A few more tries later, I obtained this puzzle on which I had a mitigated feeling, certainly due to the fact that my previous attempts (with 1-9 set) had been somewhat disastrous. So, I sent the puzzle to the other authors but I remained a bit hesitant, not totally sure if it was really good or if it was just "better than the ridiculous previous attempts".
Finally, Fred and friends told me there was no issue with this puzzle and encouraged me to keep it ; I hadn't given a look at it since its creation, so I came back to it, found the solving smoother than in my memories, and did breathe a big sigh of relief.
My two other puzzles were the Arrow Sudoku (clearly not an incredible one, but I had it created some time ago and it fitted well with the theme ; some not too bad steps to find if following the main path, though) and the Consecutive. Again, not a fantastic piece of a puzzle but I had some score to settle with consecutive puzzles and designing a themed one happened to be a good way to do it.

Unfortunately, there were a few issues due to color (but as Deb said, the use of color was an exception because of the spring theme, so it souldn't happen again) and to the fact that one or two puzzles were definitely hard to solve online. That's a pity but things rarely go as well as they were intended to, and I think that, despite of these, the balance remains positive.


I agree. By the way, I also dislike these extra conversion steps but we thought that, by giving the alphabet table, every player would quickly replace letters with numbers and that it would be a "lesser issue". Unfortunately we omitted to take into consideration the fact that online solvers couldn't replace the letters with numbers ; if only we had thought about it, we would surely have adopted the solution you propose, or something very near.

I won't extent about WaterlooMathie's last post ; answers have already been given. Let's just say that problems are always far more obvious to spot once they have happened...

All in all, a big thank you to all the players who had a try on our puzzles. I truly hope that, amongst the various grids, everybody found some puzzle which fitted him/her.

Bastien